Differential effects of aging, Alzheimer's pathology, and APOE4 on longitudinal functional connectivity and episodic memory in older adults

Abstract

Background: Both aging and Alzheimer's disease (AD) affect brain networks, with early disruptions occurring in regions involved in episodic memory. Few studies have, however, focused on distinguishing region-specific effects of AD-biomarker negative "normal" aging and early amyloid- and tau pathology on functional connectivity. Further, longitudinal studies combining imaging, biomarkers, and cognition are rare.

Methods: We assessed resting-state functional connectivity (rsFC) strength and graph measures in the episodic memory network including the medial temporal lobe (MTL), posteromedial cortex (PMC), and medial prefrontal cortex alongside cognition over two years. For this preregistered study, we included 100 older adults who were amyloid- and tau-negative using CSF and PET measurements to investigate "normal" aging, and 70 older adults who had longitudinal CSF data available to investigate functional changes related to early AD pathology. All participants were cognitively unimpaired older adults from the PREVENT-AD cohort. We used region of interest (ROI)-to-ROI bivariate correlations, graph analysis, and multiple regression models.

Results: In the amyloid- and tau-negative sample, rsFC strength within PMC, between parahippocampal cortex and inferomedial precuneus, and between posterior hippocampus and inferomedial precuneus decreased over time. Additionally, we observed a longitudinal decrease in global efficiency. Further, there was a steeper longitudinal decrease in rsFC and global efficiency with higher baseline age particularly of parahippocampal-gyrus regions. Further, lower rsFC strength within PMC was associated with poorer longitudinal episodic memory performance. In the sample with available CSF data, a steeper increase in rsFC between anterior hippocampus and superior precuneus was related to higher baseline AD pathology. Higher MTL-PMC rsFC strength was differentially associated with episodic memory trajectories depending on APOE4 genotype.

Conclusions: Our findings suggest differential effects of aging and AD pathology. Hypoconnectivity within PMC was related to aging and cognitive decline. MTL-PMC hyperconnectivity was related to early AD pathology and cognitive decline in APOE4 carriers. Future studies should investigate more diverse samples, nonetheless, our approach allowed us to identify longitudinal functional changes related to aging and early AD pathology, enhancing cross-sectional research. Hyperconnectivity has been proposed as a mechanism related to early AD pathology before, we now contribute specific functional connections to focus on in future research.

Keywords: APOE; Aging; Alzheimer’s disease; Episodic memory; FMRI; Functional connectivity.

Fig. 1

Study design and regions of interest. A Visualization of included sessions. Sample A consisted of 93 amyloid- and tau-negative (A⁻T⁻) older adults. Biomarker status was determined via CSF or PET. Participants had to be biomarker-negative via CSF at the last session of MRI and cognitive assessment we investigated, which was FU24. If only PET was available, they had to be negative at this measurement, which took place on average 6 years after baseline. If PET and CSF for sessions later than 24 months were available, both had to be below the respective threshold. Sample B consisted of 69 older adults with available longitudinal CSF measurements independent of biomarker status. B Visualization of regions of interest. Illustrations are presented in neurological view. Brain regions considered as subregions of a larger region (precuneus, medial prefrontal cortex, and medial temporal lobe) are presented in the same shade of green. Retrosplenial cortex and posterior cingulate cortex were not further divided into subregions. Medial view is based on the Brainnetome atlas, which was used for ROI definition. The coronal view is based on the aseg atlas for illustration of the hippocampus only. fMRI = functional magnetic resonance imaging. CSF = cerebrospinal fluid. PET = positron emission tomography. RBANS = Repeatable Battery for Assessment of Neuropsychological Status

Fig. 2

Longitudinal decrease in resting-state functional connectivity in amyloid- and tau-negative older adults. Illustrations are presented in neurological view. A Cluster 1 showing decrease in rsFC over one year (baseline to follow-up after 12 months). B Cluster 2 showing decrease in rsFC over one year (baseline to follow-up after 12 months). C Cluster 1 showing decrease in rsFC over two years (baseline to follow-up after 24 months). D) Cluster 2 showing decrease in rsFC over two years (baseline to follow-up after 24 months). l = left. r = right. See Table 2 for statistics

Fig. 3

Longitudinal decrease in global efficiency in amyloid- and tau- negative older adults. Illustrations are presented in neurological view. As there was no involvement of the hippocampus, the coronal view is not shown. A Decrease in episodic-network global efficiency over one year, from baseline to the follow-up after 12 months, involving the left and right superior mPFC and left anterior PRC. B Decrease in episodic-network global efficiency over two years, from baseline to the follow-up after 24 months, involving the RSC, superior mPFC, PCC, and medial PCun, each in the left and right hemisphere. C Steeper longitudinal decrease in episodic-network global efficiency over one year with higher baseline age involving the posterior PRC and EC, each in the left and right hemisphere. mPFC = medial prefrontal cortex. PRC = perirhinal cortex. RSC = retrosplenial cortex. PCC = posterior cingulate cortex. PCun = precuneus. EC = entorhinal cortex

Fig. 4

Longitudinal change in functional connectivity strength with higher baseline age and more baseline Alzheimer’s disease pathology. Illustrations are presented in neurological view. A In sample A (amyloid- and tau-negative older adults), there was a steeper longitudinal decrease in resting-state functional connectivity (rsFC) over one year with higher baseline age in the brain regions highlighted in blue. B Change in rsFC over one year was negatively associated with baseline age. The functional connection exhibiting the strongest effect (lateral parahippocampal cortex—posterior perirhinal cortex) is used for visualization. C In sample B (older adults with available longitudinal cerebrospinal fluid (CSF) measurements), there was a steeper longitudinal increase in rsFC over two years with a higher baseline p-tau₁₈₁/Aβ_1–42 ratio measured in CSF in the brain regions highlighted in red. D Change in rsFC over two years was positively associated with the p-tau₁₈₁/Aβ_1–42 ratio at baseline. The functional connection exhibiting the strongest effect (anterior hippocampus—superior precuneus) is used for visualization. l = left. r = right. rsFC = resting-state functional connectivity. PHC = parahippocampal cortex. PRC = perirhinal cortex. AD = Alzheimer’s disease. Aβ = amyloid-beta. HC = hippocampus. PCun = precuneus

Fig. 5

Association of baseline functional connectivity strength and change in episodic memory performance. Episodic memory performance measured with the RBANS delayed memory index score. A In sample A (amyloid- and tau-negative older adults), there was an association of baseline resting-state functional connectivity (rsFC) and change in episodic memory performance. There was no effect of APOE4 group. The functional connection with the strongest decrease over time, right RSC—right PCC, was investigated. N = 91. B In sample B (older adults with available longitudinal cerebrospinal fluid (CSF) measurements), there was an interaction of baseline rsFC and APOE4 group on change in episodic memory performance. N = 68. RSC = retrosplenial cortex. PCC = posterior cingulate cortex. HC = hippocampus. PCun = precuneus. RBANS = Repeatable Battery for Assessment of Neuropsychological Status. r = right